Use promote_type instead of promote_eltype in misc tests

[andreasnoack]

I think this has been wrong from the beginning but just happened to work until Julia 1.10.

There are failures on Julia 1.6 related to inference. I've identified the regression to come from version 0.3.18 of HypergeometricFunctions. It's not at all clear why it happens and Base.return_types and code_warntype disagrees

julia> Base.return_types(betalogcdf, (Rational{Int}, Rational{Int}, Rational{Int}))
1-element Vector{Any}:
 Any

julia> code_warntype(betalogcdf, (Rational{Int}, Rational{Int}, Rational{Int}))
Variables
  #self#::Core.Const(StatsFuns.betalogcdf)
  α::Rational{Int64}
  β::Rational{Int64}
  x::Rational{Int64}
  @_5::Int64
  q::Float64
  p::Float64
  _x::Rational{Int64}

Body::Float64
...

not clear what we should do here.

[codecov-commenter]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (56e1055) 63.02% compared to head (5919b62) 63.02%.
Report is 1 commits behind head on master.

Files	Patch %	Lines
src/tvpack.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #164   +/-   ##
=======================================
  Coverage   63.02%   63.02%           
=======================================
  Files          14       14           
  Lines         641      641           
=======================================
  Hits          404      404           
  Misses        237      237           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[andreasnoack]

I've put the inference check behind a check for Julia version. I don't think it's worth our time to debug an inference issue on Julia 1.6.